Datapages, Inc.Print this page

Long-Term Climate Variability in Lake Albert, East African Rift, from Geochemical Analysis of Well Cuttings

Mantaro, Jessica 1; Scholz, Christopher 1; Brown, Erik T.2
1 Earth Sciences Department, Syracuse University, Syracuse, NY.
2 Large Lakes Observatory, University of Minnesota, Duluth, MN.

Lake Albert is the northernmost of several large lakes in the western branch of the East African Rift System. Through geologic time, extensive shifts in African climate have driven variability in lake level, which is documented by several different climate proxy records. Determining the timing and extent of these lake level fluctuations can increase our understanding of past African climate variability across the region.

Long term changes in African climate are thought to be driven by events such as the closure of the Indonesian seaway (3-4 Ma) and/or uplift of central Africa through the late Cenozoic. These tectonic events and subsequent shifts in climate are likely recorded in the sediments of Lake Albert. The well cuttings, taken from an exploration well on the southeastern shore of Lake Albert, are possibly the first continuous sampling of the late-Neogene through Quaternary climate variability in continental equatorial Africa.

Samples were analyzed using the ITRAX scanning XRF at the Large Lakes Observatory, University of Minnesota at Duluth, and an ARL 8410 XRF instrument at Syracuse University. Changes in elemental abundances through time provide details on sediment chemistry and past climate change. For example, changes in the ratio of Ca/Fe can indicate changes in the abundance of carbonate accumulation through time, which provides details about fluctuations in lake level. When combined with other measurements, such as well log data (gamma ray, K, and U counts), elemental abundances can be a powerful tool in identifying horizons which may be rich in clay minerals such as Illite or Kaolinite, or horizons which may be rich in weathered silicate material.

Several multivariate statistical methods have been used to elucidate the relationship between the different XRF response variables (elements) and also between the XRF data and well log data from the same well. In a canonical correlation analysis, it is found that the XRF canonical variates contain as much as 62% of the variance in the well log data, if all of the variates are considered. In a principle component analysis, it is found that 78% of the XRF information could be explained by the first four principle components when using a minimum eigenvalue of 1.


AAPG Search and Discovery Article #90090©2009 AAPG Annual Convention and Exhibition, Denver, Colorado, June 7-10, 2009